Discrimination accuracy of models across the training and test sets
| UKB train | UKB test | WHII | |
| N | 176 611 | 44 151 | 2934 |
| Baseline model | |||
| Age only | 0.75 (0.75 to 0.75) | 0.77 (0.75 to 0.79) | 0.74 (0.69 to 0.78) |
| UKBDRS | |||
| UKBDRS | 0.79 (0.78 to 0.79) | 0.80 (0.78 to 0.82) | 0.77 (0.72 to 0.81) |
| UKBDRS-APOE | 0.81 (0.81 to 0.81) | 0.83 (0.81 to 0.84) | 0.80 (0.75 to 0.85) |
| Other risk scores | |||
| CAIDE | 0.60 (0.60 to 0.60) | 0.60 (0.58 to 0.63) | 0.69 (0.64 to 0.74) |
| DRS | 0.76 (0.76 to 0.76) | 0.77 (0.76 to 0.79) | 0.74 (0.69 to 0.78) |
| ANU-ADRI | 0.57 (0.57 to 0.57) | 0.57 (0.54 to 0.59) | 0.52 (0.45 to 0.58) |
AUCs are reported with 95% confidence intervals indicated in parentheses. 0.9% of the UKB sample had missing data for one variable of the ANU-ADRI score (BMI). Therefore, all individuals with missing data on the ANU-ADRI were first excluded before evaluating the AUC for the ANU-ADRI.
ANU-ADRI, Australian National University Alzheimer’s Disease Risk Index; AUC, area under the curve ; BMI, body mass index; CAIDE, Cardiovascular Risk Factors, Aging and Dementia; DRS, Dementia Risk Score; UKB, UK Biobank; UKBDRS, UK Biobank Dementia Risk Score; WHII, Whitehall II.